Big Idea

---
name: big-idea
description: Find the BIG IDEA of a piece — the one fresh, charged idea the whole thing hangs on, spiky enough that the reader stops and feels it. TRIGGER when a piece needs its core idea found or sharpened, when a draft reads as flat or generic, or when a headline or angle has no single idea underneath it. DO NOT TRIGGER for line polish (edit-sentences) or for writing the body once the idea is set (copywriting).
---

# Big Idea

One Process: pin the one charged idea the piece hangs on. Method distilled from the wiki's `prompts/Big Ideas.md` (400+ proven ideas) and `prompts/Big Idea Generator.md` (the volume→filter method), resolved from the wiki root per the wiki skill — `$WIKI_PATH`, default `/Users/jordan/Developer/wiki`; cite those for examples, never invent examples.

The core principle is **volume then filter**: generate far more candidate ideas than you need, then cut ruthlessly to the one that survives. Success is measured by what you eliminate.

## 1. Lock the exact domain

Name the narrow domain in the plain words a person would say — "email for SaaS trials," not "marketing." Too broad kills the impact. Test: "I do [domain]" must sound like a specific job.

## 2. Map the belief the piece fights

The strongest ideas are weapons against a bad idea, not descriptions of a good one.
- What does everyone in this exact domain hold sacred — the gospel they nod along to?
- Which of those beliefs is tired, or actually wrong?
- Name the enemy in one sentence. The big idea is the reversal that makes the reader think "wait, we've been thinking about this backwards."

## 3. Generate a set of candidates

Draft 15+ candidate ideas for what the piece hangs on. Do not self-censor — generate weak ones too. Vary the angle across:
- **Reversal:** the sacred belief turned over.
- **Binary:** X over Y, X not Y (like "Progress Over Perfection").
- **Named thing:** a metaphor or method name (like "Atomic Habits," "Purple Cow").
- **Imperative:** a single charged rule (like "Start With Why").
Pull the exact forms and worked examples from `Big Idea Generator.md`; draw real examples only from `Big Ideas.md`.

## 4. Filter ruthlessly to one

Cut every candidate that fails a check. Write the reason as you cut.
- **Oversaturation:** heard this basic idea 50 times already? Cut.
- **Anti-zoom-out:** could apply to any domain, or reads like a self-help line? Cut. Keep only what is specific to this exact domain.
- **Real enemy, not a straw man:** does the idea reverse a belief this buyer actually holds, or name a costly decision error they actually make? Cut it if the belief it reverses is one no real buyer holds — reversing a straw man reads as insight but lands on no one.
- **Problem, not feature:** is the idea an interesting way to talk about the problem, or a sexy way to talk about a feature? A feature-derived idea is cut — it is the predictable-mid-angle failure mode.
- **Conflict, not conclusion:** does the idea dramatize a conflict or reversal the buyer can feel, or does it just state a conclusion? Cut a line that only asserts the payoff.
- **Verdict inside the evidence:** the idea may dramatize the conflict as hard as it wants, but the verdict it hands down stays inside what the evidence proves. Cut the escalation or scope the idea back to what the evidence shows.
- **Pause and feel:** does it make the reader stop and feel something before they analyze it? A flat, purely intellectual idea is cut.
- **Single layer:** one idea, not several piled together.
- **Counterintuitive:** in stark contrast to what everyone else in the domain says.
- **Spiky but believable:** challenges the norm without being so extreme the reader can't buy it.
- **Belief positioning:** fill "In [domain], everyone believes [X], but the truth is [your idea]." If that sentence lands, the idea is real.

## 5. Refute, iterate, and return the survivors

Run the wide-then-filtered candidates through the shared idea process (see ideate): fresh-context adversarial reviewers refute the survivors of step 4, trivially-refutable ideas die, and each remaining idea is iterated against the criticism it drew. Return 3–5 survivors as the options — for each, the idea in a sentence, the enemy belief it reverses, the belief-positioning line, and the criticism it answered. The owner picks.

Verification: survivors are specific to the exact domain and reverse a named enemy belief, each stood a fresh adversarial review and was iterated against it, 3–5 reach the owner as options, and every cited example traces to `Big Ideas.md` — no invented examples.